A Computer-Aided Pipeline for Automatic Lung Cancer Classification on Computed Tomography Scans.

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Oncology

Background:

• Lung cancer is a leading cause of cancer mortality worldwide.
• Early detection and accurate classification of lung nodules are critical for patient survival.
• Computed Tomography (CT) scans are a primary tool for lung cancer screening.

Purpose of the Study:

• To develop and evaluate a novel computer-aided pipeline for the early diagnosis of lung cancer.
• To accurately classify detected lung nodules as benign or malignant.
• To enhance the precision of lung nodule analysis in CT scans.

Main Methods:

• A four-stage pipeline involving image preprocessing, nodule detection, feature extraction, and classification.
• Utilized a novel Lung Volume Extraction Method (LUVEM) for lung region extraction.
• Employed Circular Hough Transform (CHT) for nodule detection, Self-Organizing Maps (SOM) for segmentation, Principal Component Analysis (PCA) for feature reduction, and Probabilistic Neural Network (PNN) for classification.

Main Results:

• The proposed pipeline achieved high accuracy in classifying benign and malignant nodules: 95.91% accuracy, 97.42% sensitivity, and 94.24% specificity.
• Demonstrated effectiveness even for small-sized nodules (3-10 mm), with 94.68% accuracy.
• The LUVEM method proved significant for effective lung region extraction.

Conclusions:

• The developed computer-aided pipeline offers a promising tool for accurate and early lung cancer diagnosis.
• The system's high performance in classifying lung nodules, including small ones, supports its clinical utility.
• Integration of advanced image processing and machine learning techniques enhances diagnostic accuracy in lung cancer detection.